Instructional Design Resources

Librarians sharing cool stuff

Session B303 – Folksonomies and Tagging: Libraries & the Hive Mind October 31, 2007

Filed under: IL2007,tagging — ellenh @ 3:31 pm

Tom Reamy, Chief Knowledge Architect, KAPS Group

Cautionary quote about folksonomies – “Things fall apart; the center cannot hold; mere anarchy is loosed upon the world… The best lack all conviction, while the worst are full of passionate conviction.” -W.B. Yeats

Essentials of Folksonomies

  • folksonomy is done by users, taxonomy is done by professionals
  • basically what it comes down to is that it is metadata that users add
  • key – social mechanism for seeing others’ tags


  • Very simple to use – no need to learn a difficult classification system
  • lower cost of categorization – distribute the cost over a large population
  • Open ended – respond quickly to changes
  • Relevance – users’ own terms
  • supports a serendipitous form of browsing
  • easy to tag any type of an object – photos, docs, bookmarks
  • better than no tags at all
  • gets people excited about metadata

Disadvantages – related to quality of tags

  • they don’t work well for finding – re-finding is of marginal value
  • no structure, no conceptual relationships – flat lists do not an “onomy” make – there is only popularity
  • issues of scale – popular tags already showing a million hits
  • limited applicability – only useful for non-technical or non-specialist domains
  • either personal tags (other’s can’t find) or popularity tags – lose interesting terms. Most people can’t tag very well. Tagging is a learned skill.
  • errors – misspellings, bad compounds, etc.


  • Unwisdom of crowds, madness of crowds
  • Tyranny of the majority, popularity drowns quality, narrowing of choices, lost content
  • belief that hierarchy, taxonomy not needed

Will Social Networking make better Folksonomies?

  • Not so far – the same tags are dominating on
  • quality and popularity are very different things
  • most people don’t tag, and don’t re-tag
  • study – folksonomies follow NISO guidelines – nouns, etc. – but do they actually work to get you where you want to go? – no
  • most tags are created by heavy computer users, who love to do this stuff – the regular users and infrequent users aren’t good taggers

Flickr Facets

  • his organization analyzed flickr tag clouds – 90% of content falls into 6 different types of facets (place (40% of tags), events, dates, people, things/animals, color).
  • Subject matter of photograph was less than 1% of the tags
  • If Flickr added facets, it would be a whole lot more useable

  • tags are not facets, they are subjects
  • high-level topics – photography, news, education
  • get related terms by popularity, not conceptually
  • one type of facet stood out – “howto” “tutorial” “toread” “todo” etc.
  • popularity is not quality – dominance of computer terms, tyranny of the majority – “design” – 1 million, “interior design” – 3,909
  • top 25 – same set, slight order shift – social inertia
  • folksonomy findability – too many hits, no plurals or stemming, personal tags like “cool” “fun” and “funny” – good for social research, not for finding documents or sites
  • Folksonomies are really good for social research

Improving the Quality of Folksonomies

  • adding facets to Flickr
  • Clustering tags – taxonomy/ontology, entity extraction, populate facets and subjects, types of relationships
  • add a broad general taxonomy of most popular tags – tags as natural categories (dog, rather than “mammal” or “purebred golden retriever”)
  • evolve quality of tags and emerging structure of tags – preferred terms, ranking tags

Folksonomies and Libraries

  • Three contexts: Library Catalog, Internet Service, Enterprise (Knowledge Management) contributor
  • PennTags, Stanford – librarians find good sites to tag
  • LibraryThing – still high level concepts, not that much better at tagging, issue is the variety of terms, strange tags – 19,000 tags of “book”, combination of facets and topics, inconsistencies, redundancies

There is a lot of in-between between folksonomies and LCSH

What might work: semantic infrastructure and evolution, dynamic social rules, reduce the amount of “folk” and increase “onomy” (example: Wikipedia hiring editors, ranking articles), also can increase “folk” – not just see tags, discuss tags